| China’s Chang’E-4 probe landed on the floor of the Von Kármán crater in the South Pole-Aitken(SPA)basin on the farside of the Moon.The Yutu-2 rover is equipped with a lunar penetrating radar(LPR)system to detect the subsurface structure of the landing zone.The LPR contains two channels with center frequencies of 60 MHz(CH-1)and 500 MHz(CH-2),where the high-frequency channel CH-2contains one transmitter and two offsets receivers(CH-2A and CH-2B).The processing results of LPR data in the previous literature show the complex structure of the lunar regolith.However,the thickness of the Finsen ejecta in the landing zone is controversial,and the properties of the different strata and the regolith evolution process are not clear.Therefore,key properties that can determine the structure and properties of the regolith need to be obtained to constrain these critical issues.Since the characteristics of rocks in the regolith are strongly correlated with the regolith weathering or maturity,the analysis of the properties of the rocks in the regolith can obtain the distribution features of regolith maturity and thus reveal the regolith evolution.The analyses of the rocks in regolith rely on LPR data processing and analysis methods with high-resolution,but the existing studies on LPR data filtering and velocity inversion have not developed targeted methods based on the characteristics of LPR data.For example,the clutter interference generated by the rover and the instruments onboard it will have serious impacts on the LPR data,causing anomalous features in the radargrams and spectra,so special denoising and analysis methods need to be developed;the spatial variation of the grain size of the rocks in regolith caused by the combined effect of electromagnetic wave attenuation characteristics and heterogeneous weathering,the width of the hyperbola on the LPR radargrams will change drastically with depth.This affects the accuracy of the velocity inversion and therefore requires the targeted velocity inversion methods.In addition,the lunar regolith properties used in current studies are limited to the basic electrical parameters such as relative permittivity and loss tangent,while the polarization property of the electromagnetic wave of the lunar regolith has not been studied,which can play an important role in constraining the structure and material origin of the lunar regolith.To address these issues,this paper conducts a series of studies based on LPR CH-2 data:(1)This paper investigates the convolutional neural network(CNN)-based denoising method for LPR data.In this study,we extracted special low-frequency clutter from Chang’E-3 LPR CH-2 data using the empirical mode decomposition(EMD)method.Combined with Gaussian noise and simulated data of multiple models to train the denoising CNN,and finally obtained LPR radargram with a high signal-to-noise ratio.The comparison results with several different denoising methods show that the processing results of the proposed method present clearer and smoother events,and the stripe-like interference in the profile is more effectively suppressed.In particular,the advantage of the CNN filter is more obvious for strong noise in the deeper parts,and its processing results have lower values of image entropy.Based on the results,we have determined the layered structure in the paleo-regolith below 35 m depth.This study provides high signal-to-noise ratio LPR data for further analysis of the regolith structure.(2)This paper investigates the parameter inversion using variable-window 3D velocity spectrum based on the LPR image features.The width of the hyperbolas on the LPR profile varies drastically with depth due to a combined effect of the attenuation characteristics of electromagnetic(EM)waves and the spatial variation in the grain size of regolith caused by inhomogeneous weathering.In this study,we have developed a variable-window 3D velocity spectrum method to address this phenomenon and enhance the reliability of the EM wave velocity inversion.Parameter inversion tests of the complex regolith model show that the proposed method reduces the average inversion error by 5.42% compared to the fixed-window3 D velocity spectrum method.We then obtained the high-precision distributions of regolith EM wave velocity,relative permittivity,density,and Ti O2 and Fe O content in the landing area,which provide a basis for subsequent regolith modeling and interpretation of the regolith structure and the origin of the material.(3)This paper investigates the methods for analyzing the grain size,abundance,and distribution characteristics of rocks in regolith.In this study,a time-frequency analysis method based on variational mode decomposition(VMD)is firstly developed for the dual-peak structure in the CH-2B spectrum of the LPR data,and the dual-band data in CH-2B data were obtained and compared.The features of the lunar rock grain size in the landing area are obtained.Besides,a local unit correlation(LUC)based data similarity analysis method is proposed for analyzing the similarity of the LPR CH-2A and CH-2B data.This method is verified by numerical simulation and a detailed analysis of the parameter selection method is carried out to finally obtain the horizontal distribution features of lunar rocks in the landing area.Based on the grain size,abundance,and distribution characteristics of rocks in regolith obtained by the two methods,this paper reveals the horizontal variation characteristics of the rocks and the layer absences in the lunar regolith,and further reveals a buried pale impact crater in the landing area and the abnormally abundant rock fragments filled in the crater.Combined with the impact crater formation theory,the approximate time of formation of the buried crater and the origin and cause of the formation of the abnormally abundant rock fragments within the crater are clarified.The results also show that the fine-grained regolith at a depth of 0-12 m has a different material originating from the coarse-grained spatter layer at a depth of 12-35 m.The thickness of the Finson impact crater spatter in the landing zone can thus be determined to be~12 m.(4)This paper inverses the polarimetric scattering matrix of the lunar regolith based on the rover-turning LPR data,and the maturity distribution of the lunar regolith is obtained by the full-polarimetric target decomposition method.The method of inversion of the scattering matrix is investigated based on the LPR CH-2 data collected at the same location and different measurement angles during the rover turnings,and the validity of the method is verified by simulation tests with different models.The distribution features of the surface-like and volume scattering components of the lunar regolith were then obtained using the classical Freeman decomposition.The relationships between the two scattering components and rock size,relative permittivity,and the degree of clutter are further proposed to quantify the regolith maturity in the landing area.By analyzing the spatial distribution of regolith maturity and the special structures in the regolith,and by combining the new insights obtained from rock analysis techniques,we finally establish a fine evolution model for the shallow regolith in the landing area.In this model,we explain the differences in the spatial distribution of rocks in the coarse-grained regolith layer and the causes of spatial heterogeneity of the lunar surface,elucidate the formation process of shallow fine-grained regolith and the buried crater,and also reveal the origin of rocks in the buried crater and its inhomogeneous weathering process.This paper establishes a lunar rock analysis technique for LPR CH-2 data containing CNN filtering,variable window 3D velocity spectra,VMD,and LUC,and a lunar regolith polarization property analysis technique based on LPR data collected during the rover turnings for the first time.The two techniques reveal the evolution of the shallow lunar soil in Chang’E-4 landing area,providing both new technical support for the analysis of lunar regolith structure and properties using LPR data and new insights into the evolutionary patterns of lunar regolith. |
PDF Full Text Request